Parsing

Parses board game rulebook PDFs into Markdown using Docling, for use as source documents in boardgame_rag.py.

This step uses its own environment, separate from the RAG pipeline's — see environment.yml in this folder.

Usage

Run from the repo root (not from inside parsing/):

from parse_rulebook import parse

parse("catan.pdf", "catan")

# For scanned/image-based PDFs, enable OCR:
parse("betrayal_at_house_on_the_hill.pdf", "betrayal_at_house_on_the_hill", ocr=True)

parse() converts a single rulebook PDF to Markdown and writes it to parsed/.

file_name — PDF filename from <repo_root>/rulebooks/
out_name — base name for the output Markdown file
ocr — set True to run OCR (via RapidOCR) on scanned/image-based PDFs; defaults to False

Paths

Reads rulebooks from <repo_root>/rulebooks/
Writes parsed Markdown to <repo_root>/parsed/

The script locates the repo root via Path(__file__).resolve().parent.parent, which assumes this file stays at <repo_root>/parsing/parse_rulebook.py. If you move it, update that line accordingly.